Efficient federated learning secure aggregation method based on symmetric homomorphic encryption algorithm

The invention discloses an efficient federal learning security aggregation method based on a symmetric homomorphic encryption algorithm, and the method comprises the steps: encrypting a local gradient by a plurality of users through a public key of a central service, and then transmitting a ciphertext to an aggregation server; after aggregation, the aggregation server sends an aggregation ciphertext to the central server, and the central server decrypts the aggregation ciphertext to obtain a global gradient; according to the technical scheme provided by the invention, a homomorphic signature scheme is adopted to verify global model parameters, so that the correctness of an aggregation result is ensured, potential threats such as privacy disclosure and data forgery can be overcome in federated learning, the security and reliability of data are ensured, and the calculation and communication overhead is reduced. 本发明公开了一种基于对称同态加密算法的高效联邦学习安全聚合方法，多个用户将用中央服务的公钥加密局部梯度，然后将密文发送到聚合服务器；聚合服务器聚合后，将聚合密文发送给中央服务器，中央服务器解密后得到全局